Video: https://youtu.be/qZfy_7WJwow
16:18:55 From Mark Anderson (Portsmouth, UK) : What do mean by ‘read?’
16:19:13 From Mark Anderson (Portsmouth, UK) : Read out loud? Precis?
16:19:58 From Frode Hegland : Read to learn.
16:20:47 From Mark Anderson (Portsmouth, UK) : I’m not sure what that means in reality (as opposed to conceptually). What is the ‘AI’ doing our eyes/mind can’t?
16:21:09 From Frode Hegland : Alan, source name and link in chat here please
16:22:14 From Frode Hegland : I am not saying AI will read for us, I am saying AI can help summarise and make connections and views for us
16:22:57 From Alan Laidlaw : https://studio.ribbonfarm.com/p/the-physics-of-intelligence
16:23:57 From Dene Grigar : Eliza
16:24:37 From Pete Kaminski : With a text/knowledge power tool like ChatGPT, “reading” vs. “writing” is a bit of a false dichotomy. I use ChatGPT to play and interact with texts; sometimes summarizing or transforming, sometimes creating as a way of exploring an existing text.
16:24:48 From Mark Anderson (Portsmouth, UK) : (for later readers) re Eliza: https://en.wikipedia.org/wiki/ELIZA
16:28:30 From Dene Grigar : It has been useful in the classroom in that I use it to have my students query ChatGPT and then develop the depth on the topic from the information they reeceive
16:30:20 From Mark Anderson (Portsmouth, UK) : But, amongst the ‘stuff’ is where real insights lie. The ‘stuff’ is often the scaffolding. What is ‘stuff’ also apples differently to different readers (cultural, linguistic, experience).
16:31:27 From Mark Anderson (Portsmouth, UK) : apples → applies.
16:31:54 From Dene Grigar : Yes Mark
16:33:09 From Dene Grigar : Having my students compare a response to a query with the 100+ citations in your artlcle show them how depth occurs
16:33:38 From Frode Hegland : Daveed your mic went scratchy
16:33:47 From Dene Grigar : Hi Stephanie!
16:34:13 From Mark Anderson (Portsmouth, UK) : FWIW, that was my slow typing referring to stuff as unwanted load in the reading/writing. I don’t disagree, but I’m wary of my (in)ability to always detect/provide the good vs. stuff. ‘AI’ can certainly help.
16:34:24 From Dene Grigar : But then who knows how this technology will develop over the years. It could be more than a starting point of thought
16:34:58 From Mark Anderson (Portsmouth, UK) : Reacted to “But then who knows h…” with 👍🏻
16:37:21 From Frode Hegland : “I prompted this” ?
16:37:26 From Dene Grigar : It is also helping my students to solve issues they are having with their code.
16:37:49 From Peter Wasilko : Reacted to ““I prompted this” ?” with 👍🏻
16:39:56 From Mark Anderson (Portsmouth, UK) : For Frode, re metadata, does AI contribution add to things we may want to standardise with in metadata (until AI agents are promoted to human-equivalent status).
16:39:57 From Dene Grigar : ChatGPT says in my bio that I did my PhD at Berkeley.
16:40:34 From Dene Grigar : So the body of knowledge we have access to is not always valid
16:40:51 From vinton cerf : i have a comment on large language model usage
16:40:51 From Daveed Benjamin : I am seeing us having our own personal AI
16:41:10 From Dene Grigar : It is incumbent upon us to verify what we read
16:41:21 From Frode Hegland : Maybe also credit what other software, such as what writing tool and systems? They all influence thought. Photography without or with LightRoom?
16:42:21 From Mark Anderson (Portsmouth, UK) : @daveed, we might need discrete work and home AIs.
16:42:23 From Frode Hegland To Peter Wasilko(privately) : What is your topic, similar or different?
16:42:37 From Peter Wasilko To Frode Hegland(privately) : Similar physchological realism
16:42:47 From Mark Anderson (Portsmouth, UK) : Reacted to “Maybe also credit wh…” with 👍🏻
16:44:41 From Daveed Benjamin : Nature published an article that says science is dying because even though more papers and patents are bing published, transformative papers and patents have declined a huge amount. One might argue that we need to use AI however it makes sense to shift this.
16:44:53 From Pete Kaminski : Reacted to “Maybe also credit wh…” with 👍
16:46:20 From Jack Park : “long term memory” for ChatGPT: https://github.com/pashpashpash/vault-ai
16:46:37 From Leslie Carr : My position is that ChatGPT doesn’t even hallucinate (after Wolfgang Pauli)
16:47:28 From Jack Park : There is always the carbon footprint associated with LLM training and operation
16:47:47 From Dene Grigar : Leslie, your comment reminds me of the novel, “do androids dream of electric sheep?”
16:47:59 From Mark Anderson (Portsmouth, UK) : Re Nature article:: Might this atrophy be due to the industrialisation of academic research: publish-or-die, so much, too fast, not deep enough?. Perhaps introducing a correction in the over-production to also help.
16:48:45 From Dene Grigar : It could also be due to it indexing unverified information.
16:49:00 From Pete Kaminski : https://en.wikipedia.org/wiki/Literate_programming
16:49:51 From Matthias mprove.net : Peter Wasilko just mentioned “psychological realism”. This might be something like “cognitive adequacy” – the latter term was used during my AI education in cognition science to distinguish weather a system just produces sane human-like results, or weather it is essentially similar to the human brain and cognition process. For instance a computer playing chess is not considered cognitive adequate b/c it is a matter of brute force and plenty of backtracking (to keep it simpler) This is a totally different approach compared to a human chess player.
16:50:07 From Alan Laidlaw : for the risk theme, yes it is very possible it’ll get worse overtime. Ex: fraud model at Stripe. they made a model that was so effective that they no longer had new training data. they had to go back and make a worse model to allow error so they could continue to check the fraude model’s effectiveness.
16:50:29 From Jack Park : not open acces https://www.nature.com/articles/d41586-022-04577-5
16:51:06 From Peter Wasilko : Reacted to “https://en.wikipedia…” with 👍🏻
16:52:17 From Peter Wasilko : Here is the main Literate Programming community site where you can find citation and current tooling links: http://www.literateprogramming.com
16:52:23 From Daveed Benjamin : Open access version of Nature article https://www.nature.com/articles/s41586-022-05543-x
16:53:35 From Jack Park : Wolfram’s notebook was an early example of literate programming; these days, Jupyter is taking that to the masses.
16:54:56 From Frode Hegland : “Fashion is ephemeral. Art is eternal. Indeed what is a fashion really? A fashion is merely a form of ugliness so absolutely unbearable that we have to alter it every six months!”
16:54:59 From Frode Hegland : WIlde
16:55:56 From Daveed Benjamin : LLMs will be integrated into all tools in the next several years. It will simply thee and people will use it. I think the bigger concern is autonomous AI rather than chatbots. What does autonomous AI look like in AI? What are the risks and benefits?
16:56:46 From Alan Laidlaw : Leslie, you’ve got it. that’s what I’m most excited about.
16:56:47 From Mark Anderson (Portsmouth, UK) : !! Broken citation processes – re Les’ point about improving AIs ability to reference things.
16:56:47 From Peter Wasilko : I love that descriptor: Prototypist
16:57:20 From Frode Hegland To Alan Laidlaw(privately) : Can you pause your point? A few lighting talks in a few mins
16:57:45 From Mark Anderson (Portsmouth, UK) : A challenge is to re-imagine citations post-paper/print.
16:58:00 From Frode Hegland : Yes, visual-meta Mark! 🙂
16:58:42 From Peter Wasilko : Literate Programming opens The Black Box of what the programmer is thinking when coding and lets us capture metadata on alternative options and considerations that lead to the resulting programming being written the way it was.
16:58:43 From Mark Anderson (Portsmouth, UK) : Reacted to “Yes, visual-meta Mar…” with 😀
16:58:49 From Daveed Benjamin : Blockchain will be useful for provenance and also containment of autonomous AI
16:59:53 From Peter Wasilko : It never hurts to have a big physical Knife Switch on the data center housing any potentially runaway AI’s.
17:02:02 From Frode Hegland : (My lightning: fighting back through interaction speeds. Fluidity as Les said)
17:04:17 From vinton cerf : what i like about the notebook is that you can draw things more easily than with a mobile or a pad…
17:04:26 From Frode Hegland : Reacted to “what i like about th…” with ❤️
17:05:22 From Frode Hegland : So we need to build this! 🙂
17:10:03 From barbaratversky : Thanks, Alan, lovely way of showing that there are so many ways of creating!
17:10:39 From Mark Anderson (Portsmouth, UK) : OMG, having my homework auto-marked live 😲
17:10:46 From Leslie Carr : Alan, should we tie ChatGPT and Midjourney together to turn our narratives into visual form. A graphic fluency.
17:10:57 From Alan Laidlaw : Thank you, Barbara. A bit of a scattershot to see what lands. GPT is certainly a new kind of canvas.
17:11:37 From Matthias mprove.net : Bill Buxton once said: design sketches need a social life. Alan, your spring board reminded me of this. To me it seems to be a living source and place of inspiration where you, maybe your colleagues and eventually an AI engine might contribute pieces and snippets to the benefit of all participants.
17:12:19 From Alan Laidlaw : Leslie: yes! I’m seeing if I train Meta’s segment model to understand my qualia shorthand (to normalize it) and then ship it to GPT and see what happens.
17:13:09 From Alan Laidlaw : But there’s really so many interesting directions it can go and I haven’t really figured out what the nearest path is.
17:13:51 From Alan Laidlaw : Matthais: exactly!
17:15:48 From Jack Park : Reacted to “what i like about th…” with ❤️
17:15:53 From Mark Anderson (Portsmouth, UK) : Alan. Amen!
17:16:56 From Daveed Benjamin : I think we will have personal chatbots as well as chatbots for group conversations a la David E Smith’s The Augmented Conversation
17:17:35 From Dene Grigar : You are giving him too much coherence
17:17:41 From vinton cerf : LOL!!
17:18:45 From iPad. livia : A pre-meet chat between Alan’s GPT and mine would be amusing since I can barely process images! The worst thing anyone can say to me is, “let me show you” and then starts to put down lines and arrows!
17:20:09 From barbaratversky : Everything, or nearly everything, is a trade-off. Sometimes being very specific is useful, but sometimes ambiguity is useful. Poetry allows reinterpretation, making it beautiful and interesting each time it’s encountered. Similarly for visual art and music, we see new things each time we experience it. Ditto creating, artists and designers draw for one reason and when they reinspect their sketches, they see new relationships and interpretations. What’s more, experienced artists and designers are better at seeing new things in their sketches than less experienced. Creation is a process of focusing and broadening, of convergent and divergent thinking.
17:20:14 From Leslie Carr : We don’t live longer, it just feels like it
17:20:19 From Daveed Benjamin : ChatGPT handles visuals as inputs to prompts
17:23:31 From Alan Laidlaw : Very true Barbara. we forget that they are trade-offs. Sometimes we can simply pull out a sheet of paper to “opt-out” of a stack of trade-offs. And other times, say with highways, we can’t really opt out.
17:23:37 From Frode Hegland To Fabien Benetou(privately) : Soon…
17:24:17 From Peter Wasilko To Frode Hegland(privately) : I’d like to say a few words about how we might augment the research process — my dream library thoughts and do a micro FastCase UI Demo
17:24:42 From Frode Hegland To Peter Wasilko(privately) : Sure, right after Fabien 🙂
17:29:02 From Alan Laidlaw : Fabien is right about this point.
17:31:25 From Leslie Carr : I notice that Vint claimed that degrees were inadequate, but not that the ACM was inadequate.
17:32:56 From Theodor Nelson : Gotta go, thanks everybody.
17:33:09 From Alan Laidlaw : Thanks Ted!
17:34:53 From Leslie Carr : Apologies, I have to go too.
17:38:57 From Matthias mprove.net : Fabien said (and I agree), “Text itself is an interface” – some decades ago we referred to these as CLI – command line interfaces. Have we gone full circle?
17:40:46 From Jack Park : have to run!
17:41:57 From Frode Hegland : You are not hallucinating, only AI can halucinate..
17:42:13 From Frode Hegland : Is language also a technology?
17:42:33 From Patrick Lichty : Casey read calls the AI the new apparatus.,
17:44:03 From Alan Laidlaw : yes, language is a technology.
17:44:15 From Pete Kaminski : human language is an interesting hybrid; it’s technologized (and socialized) native capability
17:44:31 From Frode Hegland To Peter Wasilko(privately) : Not too long right? Sorry about the timing
17:45:08 From Pete Kaminski : text is definitely a technology 🙂
17:45:08 From Alan Laidlaw : Pete: great point. “technology” is a term we made up and possibly a false dichotomy.
17:45:26 From Pete Kaminski : +1 Alan
17:46:55 From Matthias mprove.net : Sir Jen Robinson made a witty comment on brit scientists: They just use their body to move their heads around. ––– This is not life. WE ARE HUMAN BEING WITH BODIES. Bodies define us, like our mind does ––– It would be an interesting thought experiment to couple chatGPT with Boston Dynamics’ robots. /caveat: I do not want to try this experiment in the wild. https://www.ted.com/speakers/sir_ken_robinson in https://www.ted.com/talks/sir_ken_robinson_do_schools_kill_creativity/c
17:46:56 From Daveed Benjamin To Frode Hegland(privately) : Frode, I have lightning on an academic overlay for the web. I understand if there is not enough time.
17:47:04 From Pete Kaminski : The Heart of Our Cities : The Urban Crisis, Diagnosis and Cure [Gruen, Victor]
17:49:23 From vinton cerf : very cool tool
17:50:06 From Pete Kaminski : https://www.fastcase.com/solo-small-firm/
17:52:58 From Dene Grigar : The Future of Text book documents what we think of important issues of our period
17:53:28 From Stephanie Strickland : How will the chat messages be handled?
17:53:58 From Daveed Benjamin : What Peter suggested for mapping connections could be implemented in an academic overlay over the web
17:56:41 From Dene Grigar : The chat can document weblinks and other ideas that come to us during more formal talks
17:56:51 From Mark Anderson (Portsmouth, UK) : Reacted to “The chat can documen…” with 👍🏻
17:56:59 From Peter Wasilko : The Heart of Our Cities: The Urban Crisis: Diagnosis and Cure by Victor Gruen
17:57:41 From Dene Grigar : I volunteer to curate the chat.
17:57:47 From Matthias mprove.net : Matthias Müller-Prove mprove@acm.org https://mprove.de
17:58:41 From Daveed Benjamin : An academic overlay for Academia
17:59:16 From barbaratversky : Frode, you might want to edit the chat; I see Dene has just offered to do so, great!
17:59:54 From Dene Grigar : Barbara, I see retaining the original format but providing a more readable version too
18:02:00 From Frode Hegland : Reacted to “Frode, you might wan…” with ❤️
18:02:47 From Dene Grigar : Coding, server space
18:02:53 From Dene Grigar : Information management
18:03:04 From Matthias mprove.net : Thank you Peter – your presented system looks much like Hans Rosling’s, just for legal cases. §§ http://blog.mprove.net/tag/rosling/
18:03:06 From Dene Grigar : Front end design, etc
18:03:07 From Alan Laidlaw : transcripts make much more sense if you have “tokens of silence” that kind of space the transcript out like a 19th-century poem. the tone of voice emerges. (this doesn’t currently exist)
18:03:14 From Peter Wasilko : LaTeX usage should be a get ed requirement!
18:03:23 From Peter Wasilko : *gen ed
18:03:27 From Frode Hegland : No… no LaTEx….
18:03:58 From Peter Wasilko : Maybe ChatGPT could author our LaTeX?
18:04:28 From Dene Grigar : I need to get to another meeting. Thank you, Frode. Bye, colleagues.
18:04:49 From Peter Wasilko : “Typeset this using LaTeX in the style of Edward Tufte.”
18:05:09 From Frode Hegland : Reacted to “Maybe ChatGPT could …” with 😂
18:06:45 From Peter Wasilko : I want to augment the next Doug Engelbart!
18:06:54 From Pete Kaminski : “Typeset this using Typst, to resemble the style of Hofstadter’s Le Ton Beau de Marot“
18:07:02 From barbaratversky : I need to leave, thanks, Frode and everyone for a fascinating discussion!
18:07:25 From Patrick Lichty : Thank you!